MCPcopy Create free account
hub / github.com/OpenGVLab/UniFormerV2 / InputDataFields

Class InputDataFields

slowfast/utils/ava_evaluation/standard_fields.py:35–97  ·  view source on GitHub ↗

Names for the input tensors. Holds the standard data field names to use for identifying input tensors. This should be used by the decoder to identify keys for the returned tensor_dict containing input tensors. And it should be used by the model to identify the tensors it needs.

Source from the content-addressed store, hash-verified

33
34
35class InputDataFields(object):
36 """Names for the input tensors.
37
38 Holds the standard data field names to use for identifying input tensors. This
39 should be used by the decoder to identify keys for the returned tensor_dict
40 containing input tensors. And it should be used by the model to identify the
41 tensors it needs.
42
43 Attributes:
44 image: image.
45 original_image: image in the original input size.
46 key: unique key corresponding to image.
47 source_id: source of the original image.
48 filename: original filename of the dataset (without common path).
49 groundtruth_image_classes: image-level class labels.
50 groundtruth_boxes: coordinates of the ground truth boxes in the image.
51 groundtruth_classes: box-level class labels.
52 groundtruth_label_types: box-level label types (e.g. explicit negative).
53 groundtruth_is_crowd: [DEPRECATED, use groundtruth_group_of instead]
54 is the groundtruth a single object or a crowd.
55 groundtruth_area: area of a groundtruth segment.
56 groundtruth_difficult: is a `difficult` object
57 groundtruth_group_of: is a `group_of` objects, e.g. multiple objects of the
58 same class, forming a connected group, where instances are heavily
59 occluding each other.
60 proposal_boxes: coordinates of object proposal boxes.
61 proposal_objectness: objectness score of each proposal.
62 groundtruth_instance_masks: ground truth instance masks.
63 groundtruth_instance_boundaries: ground truth instance boundaries.
64 groundtruth_instance_classes: instance mask-level class labels.
65 groundtruth_keypoints: ground truth keypoints.
66 groundtruth_keypoint_visibilities: ground truth keypoint visibilities.
67 groundtruth_label_scores: groundtruth label scores.
68 groundtruth_weights: groundtruth weight factor for bounding boxes.
69 num_groundtruth_boxes: number of groundtruth boxes.
70 true_image_shapes: true shapes of images in the resized images, as resized
71 images can be padded with zeros.
72 """
73
74 image = "image"
75 original_image = "original_image"
76 key = "key"
77 source_id = "source_id"
78 filename = "filename"
79 groundtruth_image_classes = "groundtruth_image_classes"
80 groundtruth_boxes = "groundtruth_boxes"
81 groundtruth_classes = "groundtruth_classes"
82 groundtruth_label_types = "groundtruth_label_types"
83 groundtruth_is_crowd = "groundtruth_is_crowd"
84 groundtruth_area = "groundtruth_area"
85 groundtruth_difficult = "groundtruth_difficult"
86 groundtruth_group_of = "groundtruth_group_of"
87 proposal_boxes = "proposal_boxes"
88 proposal_objectness = "proposal_objectness"
89 groundtruth_instance_masks = "groundtruth_instance_masks"
90 groundtruth_instance_boundaries = "groundtruth_instance_boundaries"
91 groundtruth_instance_classes = "groundtruth_instance_classes"
92 groundtruth_keypoints = "groundtruth_keypoints"

Callers

nothing calls this directly

Calls

no outgoing calls

Tested by

no test coverage detected